This invention relates to machine vision systems, and in particular to machine vision systems having application in the handling of packages.
In the shipment of parcels and packages, large truck trailers and the like must be loaded at a hub or sorting facility with a large number of packages. Conventionally, the majority of packages are parallelepipeds. Thus, such packages can conveniently be stacked in loading trailers. Such loading of trailers is conventionally accomplished manually. However, manual loading suffers from the disadvantages that the packages are frequently manually stacked in an unstable manner which may result in package shifting in transit and resulting damage. In addition, when packages are manually stacked, the space utilization within the trailer is ordinarily inefficient.
It has thus been proposed that an automated system be developed for stacking of packages within a trailer. Such an automated system would desirably include an apparatus to measure the weight and size of each package and, through appropriate algorithmic techniques, select a location on a stack for placement of each package with a high degree of stability and efficient utilization of space. Upon determination of a location, a robotic arm and gripper or hand, provided in the automated system, would place the package in the selected location.
A feedback system is required to determine whether each package, once placed by the gripper is in fact in the selected location. Such feedback information is required to avoid package collision in subsequent stacking. Machine vision systems might provide such a feedback system. However, any feedback system must determine in real time whether each package is placed properly. Thus, calculations of whether the package is properly placed must take place within the cycle time of package placement. This cycle time is desirably short, in order to provide a high throughput of packages. For example, a desirable system might have a cycle time of two seconds.
The complete determination of the position of an object on a stack, and, more specifically, the determination of the displacement of an object from a desired position, requires obtaining six offsets with respect to the desired position. These offsets include, in a three-axis coordinate system, three translation offsets due to the translation of the package along each axis, and three rotational offsets due to the rotations of the package about the three axes. Existing machine vision systems designed for inspection and sensing over the conveyor belt lack the capability of measuring six offsets from a single image. Systems, such as stereo sensing systems, have been provided which are capable of analyzing images made from different locations, but such systems require processing a large quantity of data. Processing a large quantity of data is either difficult or impossible to achieve in real time or, if possible, requires substantial expense in computational hardware.
In addition, existing machine vision systems are subject to difficulties in interpreting data as a result of uneven or unpredictable ambient lighting.
It is thus an object of the invention to provide a method and apparatus for determining the location and orientation of a rectangular object with respect to a reference location and orientation in a short period of time.
It is a further object of this invention to provide a method and apparatus for determining the location of a rectangular object with respect to a reference position that includes effective illumination of the object so that data is independent of ambient lighting conditions.
Additional objects and advantages of the present invention will become evident from the detailed description of a preferred embodiment which follows.